计算机应用 ›› 2005, Vol. 25 ›› Issue (07): 1573-1576.DOI: 10.3724/SP.J.1087.2005.01573

• 数据库技术 • 上一篇    下一篇

一种有效聚类算法的研究和实现

张永梅1,2,韩焱1,张建华2   

  1. 1.中北大学 电子信息工程系,山西 太原 030051; 2.中北大学 计算机科学与技术系,山西 太原 030051
  • 收稿日期:2005-01-01 修回日期:2005-03-03 发布日期:2005-07-01 出版日期:2005-07-01
  • 作者简介:张永梅(1969-),女,山西太原人,副教授,博士研究生,主要研究方向:数字图像处理、人工智能;韩焱(1957-),男,山西文水人,教授,博士生导师,主要研究方向:数字图像处理、电子信息的自动检测、处理和识别;张建华(1978-),男,山西太原人,助教,硕士,主要研究方向:人工智能、地理信息系统

Research and realization of an efficient clustering algorithm

ZHANG Yong-mei1,2,HAN Yan1,ZHANG Jian-hua2   

  1. 1. Department of Electronic and Information Engineering, North University of China; 2.  Department of Computer Science and Technology,  North University of China
  • Received:2005-01-01 Revised:2005-03-03 Online:2005-07-01 Published:2005-07-01

摘要:

提出了一个基于数学形态学的三维空间聚类算法。该算法通过闭合运算,将空间对象聚成类,一次完成三维空间聚类,可以快速处理非凸的、复杂的聚类形状。由于该算法基于数学形态学,所以易于实现其高性能并行算法。采用实例将算法与普通聚类算法进行了性能比较。

关键词: 聚类算法, 数学形态学, 知识发现, 空间数据挖掘

Abstract:

Based on mathematical morphology, a new algorithm of 3D spatial clustering was presented, which clustered spatial objects by closure operation. This algorithm could not only complete 3D spatial clustering at a time, and process clustering in-convex and complicated objects rapidly. On the basis of mathematical morphology, its high performance parallel algorithm was easy to realize. Experiments show that the algorithm is better than general clustering algorithms in some cases.

Key words: clustering algorithm, mathematical morphology, knowledge discovery in databases (KDD), spatial data mining

中图分类号: